Last Week's Question
The question from last week's column came from a reader named Randy,
who wanted to know two things about keeper leagues: 1) roster sizes,
and 2) the ideal number of players to keep from one season to the
next. Before plunging into particular replies to the questions that
I ask in my columns, I like to start with my own best estimate of
the consensus of those who wrote in.
This week, however, I can offer no such consensus. The replies
to Randy's question were so varied that I can only allow them
to speak for themselves. I'll feature replies from both numerical
extremes and a number of those in the middle, but it appears that
readers like Randy will simply have to sift through this information
and see what appeals to them.
I will add that some of the formats outlined below were advocated
because they make keeper leagues more manageable and/or competitive,
while others were suggested as ways of making keeper leagues truer
to the spirit of the NFL. What you want to accomplish with your
keeper league is of course a matter for you and your league mates
to determine. But whether your purpose is to mimic the NFL as
thoroughly as possible or simply to complicate things from a strategic
standpoint by incorporating a keeper component, you will almost
certainly find something useful in the responses below.
I'll start with what Jay's league does because his approach is
one that could easily be adopted by many standard redrafter leagues
without causing much stress or confusion:
We do not play in a true keeper league, but wanted to reward owners who had the insight (luck?) to draft a break-out star. We carry a 17-man roster and allow each owner to keep one player under 3 conditions: 1) he was selected in the 8th round or later; 2) he remains on a team's roster all year; and 3) he must be QB, RB, WR, or TE - no kickers or defenses. This system has allowed years in which Clinton Portis, Jeremy Shockey, and Chad Pennington were keepers. In Pennington's case, [he was kept this year by the owner who drafted him last year] knowing he had a broken hand but if the team could survive without him, he would be a great keeper this year. It also makes for a lot of projections in the later rounds.
Although the roster size in keeper leagues varies wildly from one league to the next, the most popular number of keepers was either three or "up to three," with variations such as the following, which comes from Charles.
We have a rookie draft about 4 weeks before the regular draft each season, and we've been doing the keeper format for 3 years now. In our league, you can keep up to 3 players, with each counting towards a draft spot. So if an owner decides to keep 3 players from the previous year's team, his first pick from the player pool would be in the 4th round. Anyhow, if an owner keeps his pick from the rookie draft the entire season, he can choose to keep that player next season (which would allow the owner 4 keepers instead of 3) and the rookie keeper counts as the 12th pick in the following year's draft.
Pablo wrote in to advocate the "up to three" keeper formula and to pass along some unapologetically subjective observations about various keeper formats:
In regards to keeper leagues, I am currently in leagues
that follow four different formats: 1) a Dynasty league (keep
ALL players year-to-year); 2) a 'must keep' 3-player Keeper league;
3) a league in which each owner can keep UP TO 3-players - this
makes it very interesting as those who keep less than 3 have first
crack at available free agents/rookies until they equalize at
the 3-players threshold, then the draft proceeds as usual; and
4) a redraft league. I've been involved in FF for years - including
IDP leagues - and have been a commissioner, so I can view this
issue from all sides.
While the responses of Jay, Charles and Pablo might be the most generally useful, Jim's is the one that responds most precisely to Randy's original question:
My favorite is the Dynasty league format, though as the owner
with Smokin' Ricky Williams, Michael Bennett, Stephen Davis/DeShaun
Foster, I clearly (and very painfully!) see the drawbacks of this
system. Of course injured stud players must be kept for next year,
yet that severely hampers my available roster space for waivers
Dynasty makes FF teams similar to the past in "real" sports, where
star players routinely stayed with one team for the majority of
their careers. My opinion is this makes an owner more comfortable
and attached to the team, and such owners are likely to stick
with their leagues longer. Weekly waivers & an annual draft with
order determined by record/least points scored gives sub-par teams
the ability to become competitive rapidly.
My second favorite would be the "up-to" 3-player keeper format.
This past year another owner and I kept only 1 player each (Ahman
was mine) and I was able to pick up Domanick Davis & Chad Johnson
BEFORE other owners drafted. I've gone from a miserable 5-11 team
to the playoff hunt, so this in-between version offers the ability
to very quickly improve the team while still keeping some favorite
Our league uses a very large roster, and flexible lineups, as a direct result of our 'Keeper' rule. Our owners are allowed to select 4 offensive and 4 IDP keepers on a 22-player roster, to be named 2 weeks prior to the next draft. Why so many, so late? Three main reasons. First, it cuts nearly two hours off of the draft day work (least important, unless you're running the show). Second, we draft in reverse order of the previous season's finish (see NFL draft), and with the late deadline, this arrangement allows us to make trades throughout the off-season. And lastly, flexible lineups (rules allow for a minimum lineup and then 3 offensive players, plus 2 IDP, of owner's choice) keep any one owner from cornering the market on a particular position. It all adds up to parity, and it works! Each of our 12 teams have made the playoffs, at least once, in the last three seasons.
You think a 22-player roster is big? Steve wrote in about a league with rosters almost twice as deep.
I play in an IDP league. We have 14 owners and hold 40-man rosters. When we have our cut downs in July, we keep 25 players on our rosters, with a 15-round draft, based on record (1st round has 8 lottery picks for the bottom teams). Pure and simple, this is closest to what actual NFL teams deal with every year, and we try to keep our league as close as possible to "real play". We do have one major difference; we do not have free agency, except for the period immediately following the draft up to the night before the first game. Once the season starts, the rosters stand, and because we are so large in teams and players used, there is no allowance for free agency.. Trades are allowed until the 10th week, but all teams must have 40 (43) players. Injuries are part of the game, so tough luck if you lose 4 or 5 to the I/R. When I came into the league 8 years ago, the league had just gone from 12 to 14 and I had to draft my team like expansion teams, from other rosters (first 5 picks) and the rest from the draft. This league has been around for 14 years, and while we may lose an owner every now and then, all are hardcore football fans and fantasy players. It makes for the best in fantasy football.
Steve's league may seem a little hard-core to many readers, but he was one of many who wrote in to argue that keeper leagues should come as close as possible to the actual circumstances of the NFL. Mark is another advocate of this mimetic approach:
True keeper leagues need to mirror the NFL itself. Once a player's
value becomes too high, teams have to let him go and look to younger
talent. We use an auction to select the first 10 players on our
rosters. Each owner has $100 to spend on those 10 players. The
amount spent on a player in the auction is his value. All players
selected after the initial 10 per roster are drafted and valued
at $1, which doesn't count toward an owner's $100 auction cap.
Pick-ups during the season are also valued at $1. There is no
cap during the season itself. The cap is only for drafting purposes.
If a player plays in more than half of the NFL team's games that
year, his value goes up $4 for next year's auction. If he plays
in less than half, his value goes up $2. At the next year's auction,
the players values are adjusted prior to the auction and an owner
can freeze players on his roster and deduct the new value from
the $100 he has to spend. Any players not frozen are available
in the auction and their value is based on the new bid. This places
a cap on high-priced players like Moss and Holmes, because their
value rises from year to year and it is harder to fit them and
9 other players under the $100 cap. Low value finds like Boldin
and Droughns can fit under your cap for years to come. This also
makes trading more interesting.
Feeling overwhelmed yet Randy? Well you're the one who had to go and ask the question, so take a deep breath because we have a lot more ground to cover. The following response to Randy's question was unsigned, but offers an elegant and flexible solution to the problem of how to handle keepers:
Example - You could be out of the playoff hunt and trade say Priest
Holmes ($40) to a contender, for Byron Leftwich ($3) and Kevin
Jones ($4). Holmes would cost you almost half of your cap money
($44) at next year's auction, while you could:
leagues that allow owners to keep players indefinitely, can allow
teams to become lopsided for years to come. Limiting teams to
keep 1 or 2 star players works, but it isn't very fun knowing
that you will probably never have a shot at many stars until they
are past their prime.
- Easily freeze Leftwich ($7) and Jones ($8)
- Have plenty left to freeze other veterans on your roster,
Joe Horn ($14) and Clinton Portis($18) and Javon Walker ($12)
- Drop overpriced others, Drew Bledsoe ($12) and Steven Davis
- Still have $41 to bid on 5 other players to fill out your
I am the commish for a league in its 5th season. We expanded from 8 to 10 teams this year. We have a roster of 27 players, which more than helps cover the bye weeks. We have a cap of 10 years of contracts max per year, regardless of the number of players. So you can sign 10 players to 1-year contracts, 1 player to a ten-year contract, or any combination in between. We also do not allow resigning. Once a player's contract is over, he goes back into the draft. It definitely makes team owners think in detail on who to sign and for how long for following seasons.
I like a lot of things about this idea, but I'm left with a few questions. Most importantly, I wonder when it is that I sign players to the contracts. Is it at the moment of acquisition? If I used this method in an ordinary redrafter league, for instance, would the first pick of the draft be something like, "Priest Holmes for three years"? Or would I be able to assemble my whole team first and then figure out how to distribute my ten years of contracts? Could I wait until the end of the season? What is the point of a one-year contract? Isn't it the case that standard redfrafter leagues engage players for a sort of one-year contract? Since all of these questions could easily be ironed out before the draft (or the auction, or whatever), I think this kind of keeper system could be useful for many leagues. But whereas this format allows fantasy owners to decide how many of their team-years they want to dedicate to players, Jeff's league uses a format that takes the duration of a player's actual NFL career into account:
I am the commish of a keeper league. My buddy and I decided to start up a keeper league, and we wanted a way to keep more than three or four guys. Our answer was that we would keep players in brackets based on their years in the NFL. We now keep 3 players that have played 1 season or that have played 13 or more seasons, 2 players that have played 2 seasons, 3 players that have played 3 to 12 seasons, 1 kicker, and 1 defense. We went with a 10-team league and each roster has 26 players on it, with the thought that we will have players on our team that we do not expect much from for a year or two.
Yet another approach comes from Bram, who "tried to configure a keeper league that doesn't have complicated salary base's but still has to be managed as if it did":
We use a 28-man roster. Each week, we start 1 QB, 2 RB's, 2 WR's,
1 TE, 2 DB's, 1 LB, 1 DL, 1 KR and 1 K. 12 total. Our scoring
system averages about 60 % offensive scoring and 40% special teams
and defensive. This was in line with NFL averages a few years
You getting all this Randy? The proper roster size for a keeper league is 17.
Or 22. Or 40. Or 27. And you should be able to keep 3 players. Or
1 for 10 years. Or 3 players in their 6th year. And you should have
to give up a pick. Or some auction money. Or something. But rookies
don't count. Or at least your first rookie doesn't count. Or maybe
your first two rookies don't count.
We allow teams to keep a starter at every starting spot. 12 players
as listed above. We also allow 1 franchise player that costs the
owner their first round pick in the following year's draft. We
allow 1 rookie roster spot that doesn't cost any draft picks the
following year. This encourages player development, and we are
considering allowing a second rookie keeper without penalty.
If a team doesn't keep a full starting lineup we have a pre-draft
free agency signing period that allows owners to draft veteran
players to fill any starting roster positions that were left vacant.
This happens just before the draft. Rookies have to come in through
the regular draft.
Essentially all the teams have the same money because they can
keep any players they want to keep as long as a starting roster
spot is open. Managing rookies and Franchise players really help
with the effect.
I'm kidding, of course. I think all of these ideas are great individually.
But lumped into a single column such as this, they are starting
to make my head spin. So I better wrap this up before my eyes get
stuck this way . . .
Ah, here's a response from Casey that speaks to roster size generally
and doesn't turn my brain to mush:
The formula that is most fair is 2 times your starting lineup minus one or two players. That way you are at least forced to make some tough calls. Anything more and you wont see many trades, pickups, or interest from teams that start slow.
Casey goes on to reiterate Randy's question (which I'm not sure I've done much to answer one way or another):
I have tried to convert our league to a keeper for a couple of years,
but the very question of how many players to keep prevents the
transition from happening. People who won last year want as many
as they can get, and the bottom feeders want only 1 or 2. Since
we are a league that uses an auction, I thought we should allow
as many keepers as you want if you can afford to take a cap hit
equivalent to where the player finished in respect to the highest
salary at that position.
It sounds like Casey is on his way to answering his own question, but he might find this suggestion from Dave's league helpful:
One of our GMs suggested keeping as many as you want at 1.5 times
what you paid for him last year. Interesting, but the obvious
problem is what to do with waiver pickups?
In my 12-team, auction keeper league, we have 16-man rosters (8 starters,
8 reserves) plus an IR. There's a $100 salary cap that's only
in effect for the draft. Each year that you keep a player, their
salary increases $5. This negates the need for a strict limit
on the number of keepers. We just let the market determine how
many players an owner can afford to keep. Typically about 25-30
players are held over from year to year. Some teams will keep
3 or 4, others keep even fewer or occasionally none.
Increasing the cost of players works in auction leagues, but since redrafter leagues are still more common, I would be remiss if I neglected to mention the very popular method of having keepers increase in what we can think of as "draft-round value" from one year to the next. A lot of folks wrote in about this method, but the response from Mark explains it in the greatest detail:
I'm in another league that's basically the same thing except the
salary increase is much more severe. That league has a $50 salary
cap, but the salary increase is still $5. That makes it much harder
to keep players. So by adjusting the salary penalty, a league
can pretty much control how many players can be kept.
We've actually thought about adjusting the salary increase by
position. As it is now, no kickers or defenses are ever carried
over and TEs have to reach super stardom before they're worth
the extra cost. I haven't come up with an equitable solution yet,
but we'll probably do something like that down the road.
Several years ago I was trying to come up with an idea
for rewarding drafting an up-and-coming player. Our draft goes
19 rounds, and any player taken in the 12th through 19th rounds
can be retained the following year at a cost of the draft slot
that is 2 rounds higher than the round in which that player was
originally drafted. You can not cut or trade that player during
the year in order to retain for the next year, and you can only
retain a player for 2 more years after you initially draft him.
He moves up 2 draft spots each year, but you can't retain once
his draft position is less than 12. For example, this year Ricky
Williams was drafted in the 12th round. If he isn't cut or traded,
he can be that owner's 10th round pick next year, but the year
after that that he goes into the regular draft. This makes the
draft exciting once you hit the 12th round and with a serpentine
draft the player with the last draft pick in the 1st round gets
the 1st draft pick in the 12th round. You can only retain 3 players
and have to announce the retained players before the draft order
is selected for the year.
Although most of those who wrote in advocated keeping only a fraction of their
fantasy teams from one season to the next, Dan wrote in to suggest
that, as a general rule, it's a good idea for teams to keep roughly
half of their rosters from one year to the next, though I suspect
this is only true for leagues that have gone through a lengthy maturation
Players like Priest Holmes, Deuce McAllister, Shawn Alexander,
Kurt Warner, Mark Bulger, Santana Moss, Javon Walker, Andre Johnson
have been drafted in the later rounds in past years before they
hit stardom. Our system offers a big reward for taking a shot
at a player and having to keep him on your roster for a year without
The roster size we've decided on this year is 35, starting 22 players
each week (11 on O and 11 on D). This gives pretty much each crucial
position a back up. For future prospects and deep sleepers we
also have included a Taxi squad consisting of 5 players.
So what's the right answer? I have no idea. All I can say is, "You pays your money
and you takes your choice." Have fun choosing.
As for the number of keepers to keep every season, it all depends,
of course, on your roster size, but as a general rule - think
half. We keep 18 players out of our 35. This still gives the coming
draft enough players to keep it interesting for about ten rounds.
This Week's Question
A reader named Bradley wrote in earlier this season to ask about
the fairest way to handle trades in a league. Most leagues handle
trades in one of two ways. The first is to have commissioners approve
or veto trades (usually with a league being able to override a commissioner's
decision provided enough owners disagree with him). The second is
to have the owners themselves vote to approve or block proposed
trades. Bradley's league uses the second method, but it appears
to be causing a problem:
Regarding trades, our 10-team league uses a voting approval process
where accepted trades are subject to a 1-day voting period. Votes
may be cast by the other eight members not involved, where at most
3 objections are allowed. Trades receiving 4 or more (50% or more
of voters) objections are rejected. We have had several trades rejected
this year, and this has made the traders mad. I would love to hear
your readers' thoughts.
I'm not sure how many people would agree with me, but I think that
generally the only reason to block a trade is because the parties
involve are suspected of colluding in order to stack one team at
the expense of another. It's not right for me to vote against a
trade just because I think that it makes my opponent's team better
the week before I have to face him. However, it seems that the system
employed by Bradley's league could easily lead to just this sort
of scenario. What motivates owners to trade in the first place?
Obviously, their incentive to trade is to make their teams better.
In a good trade, there isn't really a winner or a loser; both parties
benefit. If I have depth at running back and you have depth at receiver,
we can make a trade that genuinely improves both our teams. The
trade therefore helps both of us, but it might be intimidating to
the owners who were looking forward to playing us before I acquired
a decent receiver from you or you acquired a decent running back
from me. They vote to block the trade just because we would both
be easier to defeat if the trade weren't allowed to go through.
So what is a league to do in a situation such as this? Like Bradley,
I am interested to hear the thoughts of other readers.
Last Man Standing Picks (Courtesy of Matt)
Trap Game(s): Chicago at NY Giants:
Okay, so the trap games I've picked this season haven't really turned
out to be trap games. What's important is that these games could
have been. That being said, which game, or games, are potential
trouble this week? In looking at the schedule, the Jets at Buffalo
might be, or San Francisco at Seattle. But the game that stands
out for me and the betting community is the Chicago Bears at the
New York Football Giants.
The Giants just demolished Minnesota ,and while they might not be
looking ahead to their next quality opponent, the team may not completely
listen to Coach Coughlin about the possible letdown. This team will
be at home and think that the Bears are not that good. What they
need to remember is that any team can be beaten on any given Sunday.
#3: Indianapolis over Minnesota (4-3
As we get further into the season, it becomes very important that
you make picks not only for this week but also strategic picks for
later in the season. This is one of the weeks that you can take
advantage of not choosing a good team like the Colts earlier in
the season. Minnesota is banged up at receiver and while Kelly Campbell
is fast and Nate Burleson is solid, neither is Randy Moss. If the
Colts can make the Vikings one-dimensional, this game should be
over as early as last week's game. However, Tony Dungy's defense
that was so productive last year is now rated almost dead last,
giving up 45 points last week to the rejuvenated Chiefs (see below).
If the Vikings stay close and can run a balanced offense. Look for
Indy to lose 3 in a row.
#2: San Diego over New Orleans (5-2 This
Everyone will be jumping on the bandwagon of the Chargers after
last week. LaDainian Tomlinson should have a good day against a
Saint defense that has been anything but good. On the other side
of the ball, while San Diego is not a defensive juggernaut, they
proved that they should be more than enough to shut down Aaron Brooks
and company. According to tapes watched by the Saints coaching staff,
Deuce McAllister is not as explosive as he was the last two seasons.
Combine that with a receiving corps that is only adequate and that
becomes the formula for a loss. If the Chargers get up early at
home, look for them to sit their starters in the fourth quarter.
#1: Kansas City over Tampa Bay (6-1 This
Talk about a change in fortunes. Kansas City is just piling up the
points against teams, and while the Bucs are known for defense,
they will be hard pressed to stop the tandem of Holmes, Gonzalez
and Green. This game will be close. With Griese throwing to Clayton
all day and Pittman mixing it up, but this may be the week to pick
the Chiefs if you have been avoiding them lately.
For responses to this week's fantasy question or to share your
LMS picks, please email
me no later than 10 a.m. EST on Wednesdays during the football